ROLE OF MICRORNA AND THEIR TARGET GENES IN PATHOGENESIS OF GASTRIC CANCER

(1)

1

LITHUANIAN UNIVERSITY OF HEALTH SCIENCES MEDICAL ACADEMY

Violeta Šaltenienė

ROLE OF MICRORNA AND THEIR

TARGET GENES IN PATHOGENESIS

OF GASTRIC CANCER

Doctoral Dissertation Biomedical Sciences,

Biology (01B)

(2)

2

Dissertation has been prepared at the Laboratory of Clinical and Molecular Gastroenterology of Institute for Digestive System Research of Medical Academy of the Lithuanian University of Health Sciences during the period of 2012–2017.

Scientific Supervisor:

Prof. Dr. Gediminas Kiudelis (Lithuanian University of Health Sciences, Biomedical Sciences, Biology – 01B)

Dissertation is defended at the Biology Research Council of Medical Academy of the Lithuanian University of Health Sciences

Chairperson

Prof. Dr. Žilvinas Dambrauskas (Lithuanian University of Health Sciences, Medical Academy, Biomedical Sciences, Biology – 01B)

Members

Dr. Diana Žaliaduonytė-Pekšienė (Lithuanian University of Health Sciences, Medical Academy, Biomedical Sciences, Medicine – 06B) Prof. Habil. Dr. Angelija Valančiūtė (Lithuanian University of Health Sciences, Medical Academy, Biomedical Sciences, Biology – 01B) Prof. Dr. Edita Sužiedėlienė (Vilnius University, Biomedical Sciences, Biology – 01B)

Assoc. Prof. Dr. Aija Linē (University of Latvia, Biomedical Sciences, Biology – 01B)

Dissertation will be defended at the open session of the Biology Research Council of Lithuanian University of Health Sciences on the August 31st, 2017 at 2 p.m. in 204 auditorium of “Santaka” Valley Centre for the Advanced Pharmaceutical and Health Technologies of Lithuanian University of Health Sciences.

(3)

3

LIETUVOS SVEIKATOS MOKLSŲ UNIVERSITETAS MEDICINOS AKADEMIJA

Violeta Šaltenienė

MIKRO RNR IR JŲ REGULIUOJAMŲ

GENŲ VAIDMUO SKRANDŽIO

VĖŽIO PATOGENEZĖJE

Daktaro disertacija Biomedicinos mokslai,

Biologija (01B)

(4)

4

Disertacija rengta 2012–2017 metais Lietuvos sveikatos mokslų universitete, Medicinos akademijos, Virškinimo sistemos tyrimų instituto, Klinikinės ir molekulinės gastroenterologijos laboratorijoje.

Mokslinis vadovas

prof. dr. Gediminas Kiudelis (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, biologija – 01B)

Disertacija ginama Lietuvos sveikatos mokslų universiteto Medicinos akademijos Biologijos mokslo krypties taryboje:

Pirmininkas

prof. dr. Žilvinas Dambrauskas (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, biomedicinos mokslai, biologija – 01B)

Nariai

dr. Diana Žaliaduonytė-Pekšienė (Lietuvos sveikatos mokslų univer-sitetas, Medicinos akademija, biomedicinos mokslai, medicina – 06B) prof. habil. dr. Angelija Valančiūtė (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, biomedicinos mokslai, biologija – 01B)

prof. dr. Edita Sužiedėlienė (Vilniaus universitetas, biomedicinos moks-lai, biologija – 01B)

doc. dr. Aija Linē (Latvijos universitetas, biomedicinos mokslai, biolo-gija – 01B)

Disertacija bus ginama viešame biologijos mokslo krypties tarybos posėdyje 2017 m. rugpjūčio 31 d. 14 val. Lietuvos sveikatos mokslų universiteto Naujausių farmacijos ir sveikatos technologijų centro 204 auditorijoje.

(5)

5

ABBREVIATIONS

AG – Atrophic gastritis

Ath-miR – Arabidopsis thaliana micro ribonucleic acid

CagA – Cytotoxin-Associated Gene A (of Helicobacter pylori)

CAV1 – Caveolin 1

CDH1 – E-cadherin gene CIN – Chromosomal instability DNA – Deoxyribonucleic acid EBV – Epstein-Barr Virus EREG – Epiregulin

FAT4 – FAT Atypical Cadherin 4 FDR – False Discovery Rate

GC – Gastric cancer

GCadj – Gastric cancer adjacent

GNCA – Gastric non-cardia adenocarcinoma GTP – Guanosine-5'-triphosphate

H. pylori – Helicobacter pylori

Hsa-miR – Homo sapiens micro ribonucleic acid IRF1 – Interferon Regulatory Factor 1 JAK2 – Janus kinase 2

microRNA, miRNA – Micro ribonucleic acid

MLH1 – MutL Homolog 1 is a Protein Coding gene mRNA – Messenger ribonucleic acid

OR – Odds ratio

PDCD1LG2 – Programmed Cell Death 1 Ligand 2 PD-L1 – Programmed death-ligand 1

PIK3CA – Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha

PTEN – Phosphatase and Tensin Homolog

qRT-PCR – Quantitative real-time polymerase chain reaction RHO – (Rhodopsin) is a Protein Coding gene

RHOA – (Ras Homolog Family Member A) is a Protein Coding gene RNA – Ribonucleic acid

Small RNA-seq – Small RNA sequencing

SNP – Single-nucleotide polymorphism TLDA – TaqMan low density array

TNM – TNM Classification of Malignant Tumours TP53 – Tumor protein p53

TXNIP – Thioredoxin Interacting Protein

vacA – Vacuolating cytotoxin A (of Helicobacter pylori) WHO – World Health Organization

(8)

8

INTRODUCTION

Gastric cancer (GC) is a common malignancy and a leading cause of cancer related death in the world [1]. According to GLOBOCAN 2012, 952,000 new cases of GC were estimated to have occurred in 2012 (6.8% of the total cancer burden), making it the fifth most common malignancy in the world after lung, breast, colorectal and prostate cancers. Whereas, according to Lithuanian Cancer Registry 2012, 947 new cases of GC were diagnosed in 2012 (5% of the total) and it was the sixth most common cancer by morbidity, but it was the second leading cause of death in both sexes in Lithuania (8% of total), after lung cancer. Mortality rates are notably high because, in most cases, the disease is diagnosed at advanced stages when the treatment is likely to fail. The overall 5-year survival rate of patients with GC ranges from 10% to 30% [2]. Therefore, one of the major focuses in GC research is the evaluation of potential molecular biomarkers that could be used for early non-invasive diagnostics of this malignancy.

Both environmental and genetic factors play a role in etiology of GC; however, as in most cancers, pathogenetic mechanisms in GC are still not fully understood. Demographic and environmental risk factors for GC include older age, male sex, family history, tobacco smoking, H. pylori infection and obesity [3]. Nevertheless, the exact mechanisms of malignant transformation from H. pylori infection to chronic atrophic gastritis (AG), intestinal metaplasia and GC is still poorly understood [4]. Important studies have shown that miRNAs are already deregulated in early stages of gastric carcinogenesis including H. pylori gastritis and premalignant stages of gastric atrophy and intestinal metaplasia [5, 6].

In recent years, different studies, including genome-wide association studies (GWAS), examined genetic risk factors for GC. The discovery of microRNAs (miRNAs) has opened new opportunities for understanding of pathophysiology and molecular biology of GC [7]. Small non-coding miRNA molecules (approximately 18–25 nucleotides) regulate gene expression through sequence-specific pairing with the target messenger RNA (mRNA) and inhibition of its translation [8] and influence the development of diseases [9], especially cancer [10, 11]. Furthermore, miRNAs have been shown to have a diagnostic or prognostic role and even potential clinical implications for targeted gene therapy in cancer patients [12, 13]. Volinia et al. have provided one of the first miRNA expression profiles in GC tissue showing a specific deregulation pattern [14]. Further studies have also reported significant deregulation of miRNAs belonging to hsa-miR-17, hsa-miR-21, hsa-miR-223, hsa-miR-135 and many other families. Among reported studies

(9)

9

in GC tissues, hsa-miR-21, hsa-miR-25, hsa-miR-92, hsa-miR-223 were the most consistently up-regulated miRNAs, while 375 and hsa-miR-148 were the most consistently down-regulated [15–17].

Other studies have revealed that certain single-nucleotide polymor-phisms (SNPs) of miRNA encoding genes may alter miRNA expression and influence cancer development [9, 18]. Moreover, genetic variations within miRNA binding sites affect the miRNA–mRNA interaction. SNPs within a miRNA target can reinforce, weaken or disrupt the binding with miRNAs and change the expression of mRNA targets [19–21].

Target-gene identification may help to understand the function of miRNAs. This process is challenging because miRNAs bind to their target mRNAs over a short sequence and the effectivity of targeting are not completely understood. Computational and experimental approaches to the identification of miRNA-regulated genes are applied [20].

MiRNAs are considered to be a promising candidate for clinical diag-nosis of many malignant diseases, but this requires more research [23, 24].

(10)

10

1. AIM AND OBJECTIVES

The aim of this study was to investigate microRNA profile and the role of their target gene in gastric cancer pathogenesis and to assess the relevance of these biomarkers for non-invasive diagnosis of the malignant disease.

1.1. Objectives of the study

1. To determine association between single nucleotide polymorphisms in the predicted microRNA target genes IL12B (rs1368439), INSR

(rs1051690), CCND1 (rs7177) and IL10 (rs3024498) and the presence

of gastric cancer in European population.

2. To establish microRNA expression profile in tissue of patients with gastric cancer and atrophic gastritis.

3. To assess the expression of selected microRNAs in plasma of patients with gastric cancer and to evaluate their suitability for non-invasive diagnostic.

4. To evaluate expression differences of the predicted microRNA target genes by experimentally inhibiting hsa-miR-20b and increasing amount of hsa-miR-451a and hsa-miR-1468 in gastric cancer cell lines.

1.2. The novelty and relevance of the work

Study provides novel evidence on: (i) SNPs in miRNA binding sites of target-gene and GC, (ii) miRNA profile in gastric cancerous and precancerous conditions, as well as their suitability for non-invasive diagnostic, (iii) po-tential target-genes for miRNA. These findings provide a more detailed molecular understanding of GC and tumorgenesis. Moreover, this data might serve not only for the future scientific works, but are highly significant for diagnostics and further identification of new targets and development of novel therapeutics.

(11)

11

2. REVIEW OF LITERATURE

2.1. Epidemiology of gastric cancer

Gastric cancer (GC) is a common malignancy and a leading cause of cancer related death in the world [1]. According to GLOBOCAN 2012, 952,000 new cases of GC were estimated to have occurred in 2012 (6.8% of the total cancer burden), making it the fifth most common malignancy in the world, after lung, breast, colorectal and prostate cancers. Whereas, according to Lithuanian Cancer Registry 2012, 947 new cases of GC were diagnosed in 2012 (5% of the total) and it was the sixth most common cancer by morbidity, but it was the second leading cause of death in both sexes in Lithuania (8% of total), after lung cancer. Mortality rates are notably high because, in most cases, the disease is diagnosed at advanced stages when the treatment is likely to fail. The overall 5-year survival rate of patients with GC ranges from 10% to 30% [2].

2.1.1. Classification and histology of gastric cancer

According World Health Organization (WHO), gastric tumors are histolo-gicaly classified into epithelial and non-epithelial tumors, as shown in Table 2.1 [23].

About 90% of cancers of the stomach are adenocarcinomas and they deve-lops from the stomach mucosa, usually maintaining glandular differentiation [24]. Other less common tumors of the stomach are lymphomas (4% of total), gastrointestinal tumors, carcinoid tumor and others types of cancer, such as squamous cell carcinoma, small cell carcinoma, and leiomyosarcoma, can also start in the stomach, but these cancers are very rare.

The most frequent site of sub-cardial GC is the distal stomach, i.e. the antro-pyloric region. Carcinomas in the body or the corpus of the stomach are typically located along the greater or lesser curvature [23].

(12)

12

Table 2.1. Classification of gastric tumors according WHO [23].

Epitelial tumors Non-epithelial tumors

• Intraepithelial neoplasia – Adenoma • Carcinoma − Adenocarcinoma  intestinal type  diffuse type − Papillary adenocarcinoma − Tubular adenocarcinoma − Mucinous adenocarcinoma − Signet-ring cell carcinoma − Adenosquamous carcinoma − Squamous cell carcinoma − Small cell carcinoma − Undifferentiated carcinoma − Others

• Carcinoid (well differentiated endocrine neoplasm)

• Leiomyoma • Schwannoma • Granular cell tumor • Glomus tumor • Leiomyosarcoma • GI stromal tumor

− benign

− uncertain malignant potential − malignant

• Kaposi sarcoma • Malignant lymphomas

− Marginal zone B-cell lymphoma of MALT-type

− Mantle cell lymphoma

− Diffuse large B-cell lymphoma − Others

• Secondary tumors

Adenocarcinomas are classified according to histology and location. In the traditional Laurén classification, GC is divided into two types: intestinal and diffuse types. Intestinal type is more common than diffuse [27–29]. These two types seem to follow different precancerous processes and show clinical and epidemiologic differences, H. pylori infection is the strongest risk factor for the development of most tumors of both types [27]. The intestinal type is more common in males, blacks, and older age [28]. Diffuse carcinoma tends to affect younger individuals, mainly females; it frequently has hereditary character-ristics, perhaps modulated by environmental influences [32, 33].

The consortium of The Cancer Genome Atlas (TCGA) described mole-cular classification of GC and suggested four subtypes:

• chromosomal instability (CIN) type;

• genomically stable (GS) tumors (near-diploid type); • EBV-positive tumors;

• microsatellite instability (MSI)-positive tumors; And every of them is described:

The CIN tumors have been associated with marked aneuploidy and focal amplification of receptor tyrosine kinases, as well as mutations of TP53.

The GS type has been associated with diffuse tumors, mutations of CDH1 and RHOA, or fusions involving RHO family GTPase-activating proteins.

(13)

13

The EBV-positive tumors have been correlated with PIK3CA mutations; high levels of DNA hypermethylation; and amplification of JAK2, PD-L1, and

PDCD1LG2.

The MSI tumors display characteristic hypermutation phenotype and downregulation of MLH1 gene (Fig. 2.1) [31].

Fig. 2.1. The four molecular subtypes of gastric cancer.

Distribution of molecular subtypes in tumors obtained from distinct regions of the stomach is represented by inset charts [31].

2.1.2. Risk factors of gastric cancer

GC is a genetically heterogeneous tumor with multifactorial etiologies, associated with environmental and genetic factors.

2.1.3. Environmental Risk Factors

• Gender and ethnicity. GC is more common in men than in woman [28] .

• Age: Risk increases with age, GC is extremely rare below the age of 30 [23].

• Where a person lives: Worldwide, stomach cancer is more common in Eastern Asia, Central and Eastern Europe, Central and South

(14)

14

America. Meanwhile GC is less common in North America, Austra-lia, Africa, Micronesia (GLOBOCAN, 2012).

• Diet: Being overweight or obese increasing the risk of GC. The risk of GC increase when using smoked foods, salted fish and meats, and pickled vegetables. An adequate intake of fresh fruits and vegetables lowers the risk [32], but one new meta-analysis have shown no significant association between GC risk and flavonoid intake (OR = 0.88; 95% CI = 0.74–1.04, I2 = 63.6%) for a comparison of the

highest to the lowest category of intake [33].

• Earlier stomach surgery: The risk of gastric carcinoma increases 5– 10 years after gastric surgery, especially when the Bilroth II opera-tion, which increases bile reflux, was performed [23].

• Family history: In most studies, the familial relative risk is appro-ximately three-fold, which is larger than that observed for most other adult forms of solid cancer, with the exception of ovarian cancer. In India, Korea and Turkey, much higher relative risks have been reported [34].

• Physical activity: the risk of GC is lower among the most physically active people as compared with the least physically active people [35].

• Helicobacter pylori. The most important development in the epidemiology of adenocarcinoma is the recognition of its association with H. pylori infection. Prolonged precancerous process, such as chronic gastritis, multifocal atrophy, intestinal metaplasia, and intraepithelial neoplasia, however, H. pylori is genetically heteroge-neous, and all strains may not play the same role in the development of malignancy. There are two main virulence factors which are implicated in the progression and the severity of GC, these are Cytotoxin-Associated Gene A (CagA) and Vacuolating Cytotoxin A (VacA) which are injected and secreted by H. Pylori, respectively [39, 40].

2.1.4. Genetic Risk factors

Less than 3% of GC cases arise as a result of inherited syndromes and acquired genetic factors contributing to sporadic GC, which have been systematized and described in a review of Mairi H. McLean and Emad M. El-Omar [38].

(15)

15 2.1.5. Hereditary factors

• Hereditary diffuse GC – autosomal dominant condition is associated most frequently with a heterozygous germline mutation in CDH1 gene.; CDH1 germline mutation and potential catenin family genes implicated;

• Gastric adenocarcinoma and proximal polyposis of the stomach – implicated genes are unknown;

• Hereditary nonpolyposis colorectal cancer – mismatch repair genes such as MSH2 or MLH1 are implicated;

• Li-Fraumeni syndrome (rare) – TP53 implicated; • Peutz-Jeghers syndrome (rare) – STK11 implicated;

• Familial adenomatous polyposis (rare) – APC implicated Acquired genetic factors contributing to sporadic GC.

2.1.6. Acquired genetic factors

• Chromosomal instability, such as aneuploidy, chromosomal translo-cation, amplifitranslo-cation, deletion and loss of heterozygosity;

• Fusion genes, for example, SLCIA2 (SLCIA2–CD44 fusion protein) and ROS1 (SLC34A2–ROS1);

• Microsatellite instability, for example, hyper-methylation of promoter of mismatch repair genes, primarily MSH1;

• SNPs identified using the candidate gene approach (IL1B-31*C, IL1RN*2/*2, IL-10, TNF, IL-17A-187*A);

• SNPs identified using genome wide association studies (PSCA rs2976392*A, MUC1 rs4072037*A);

• Exome sequencing (high-throughput studies) identified driver genes such as FAT4, ARID1A and RHOA;

• Copy number profile sequencing (high-throughput studies) revealed associations with components from RTK/RAS signaling pathway, e.g., FGFR2;

• Changes in miRNA profile; e.g., overexpression of hsa-miR-181a influencing target-gene KLF6 or loss of hsa-miR-449 resulting in reduced expression of tumor suppressor genes such as TP53.

2.1.7. Atrophic gastritis

Invasive gastric carcinoma is preceded by a cascade of precancerous lesions. The distal GC can be of two types, intestinal and diffuse, each following different developmental pathways. For intestinal type GC starts from uncontrolled gastric inflammation, which may lead to mucosal atrophy and

(16)

16

hypochlorhydria and increase the risk for intestinal metaplasia, dysplasia and finally cancer [39]. Little is known about the development of diffuse type GC, however it is recognized that H. pylori and inflammation may also play a signifycant role. The first step causing histologic changes in gastric tissue is acute inflammation leading to (i) non-atrophic chronic gastritis or (ii) advance to multifocal atrophic gastritis (MAG) and further proceeding to precancerous cascade: (iii) intestinal metaplasia (first “complete” and then “incomplete”), (iv) dysplasia (first low grade and then high grade (equivalent to “carcinoma in

situ”)) and finally development of GC, which is thought to be associated with

degradation of the intercellular matrix [40] (Fig. 2.2).

Fig. 2.2. Precancerous cascade of gastric cancer

(MALT-mucosa-associated lymphoid tissue) [41].

2.1.8. Diagnosis of gastric cancer

Screening is testing for cancer in people without symptoms. In countries such as Japan, where GC is very common, mass screening of the population has helped to find many cases at an early, curable stage [42]. Studies in the United States have not found that routine screening in people at average risk of GC is useful, because this disease is not that common.

(17)

17

The following tests can help diagnose GC: complete blood count (CBC) to check for anaemia or Stool test to check for blood in the stools. The main test for GC is upper endoscopy, but it is usually performed when people have some signs or symptoms of stomach diseases. Unfortunately most patients are asymptomatic in the early stage of GC [43]. If an abnormal-looking area is seen on endoscopy, the only one way to tell for sure if it is really cancer is by doing a biopsy. The biopsy sample may be tested in 2 different ways:

Immunohistochemistry (IHC): in this test, special antibodies that stick to the HER2/neu protein are applied to the sample, which cause cells to change color if many copies are present. This color change can be seen under a microscope. The test results are reported as 0, 1+, 2+, or 3+.

Fluorescent in situ hybridization (FISH): This test uses fluorescent pieces of DNA that specifically stick to copies of the HER2/neu gene in cells, which can then be counted under a fluorescent microscope.

Other imaging tests could be used, such us computed tomography scan, magnetic resonance imaging scan, positron emission tomogramphy scan, sound waves scan and X-ray. But not all methods are readily available, some of them are expensive or invasive.

2.2. MicroRNA

More and more data suggest that small non-coding RNAs such as microRNAs (miRNAs) can be utilized as potential biomarkers for the diagnosis and prognosis of a variety of diseases such as Type-II diabetes, cardiovascular disease, neurological disorders, cancer and other.

2.2.1. MiRNA biogenesis

MiRNAs are ∼ 22 nucleotides in length, small non-coding RNAs that function as guide molecules in RNA silencing [44]. Since 1993 a huge number of small RNA classes have been identified, including miRNAs, small interfering RNAs (siRNAs) and Piwi-interacting RNAs (piRNAs). These classes differ in their biogenesis, modes of target regulation and in the biological pathways they regulate [45].

The human genome encodes for over 1,800 miRNAs and they regulate about 60% of the human protein-coding genes. Moreover, single miRNA may target hundreds of mRNA [49, 50]. They play a role on the organ development, differentiation and function, as well as tumor formation, progression, invasion and metastasis.

(18)

18

MiRNA genes are scattered among chromosomes in humans, except for the Y chromosome and they are located within different genomic regions (intragenic or intergenic) (Fig. 2.3). The primary miRNA transcripts are trans-cribed as precursor molecules called pri-miRNAs by RNAPII. Pri-miRNAs are derived either from annotated transcripts (as the introns of protein coding genes, the exons of noncoding genes, or the introns of noncoding genes) or from intergenic regions within the genome and can encode a single or multiple miRNAs [48].

RNase Drosha further processes pri-miRNA into 70- to 100-nt hairpin-shaped precursors, called pre-miRNA, which are exported from the nucleus by exportin 5. In the cytoplasm, the pre-miRNA is cleaved by Dicer into a miRNA:miRNA* duplex. Both arms of the mature product from a hairpin have the potential to function as miRNAs, however which of them will be subsequently loaded into the RISC complex depends on tissue or develop-mental stage of the cell [49]. MiRNA arm switching is likely to be associated with a change in miRNA function, i.e. a single miRNA precursor encodes two mature miRNAs, with distinct targeting properties [49].

Assembled into RISC, mature miRNA negatively regulates gene expres-sion by either translational represexpres-sion or mRNA degradation, which is depen-dent on sequence complementarity between the miRNA and the target mRNA. A role of miRNA depends on its target-genes, i.e. miRNA could be an oncogene or a tumor suppressor gene. A recent computational analysis showed that miRNAs from the same precursor not only target different genes, but also generally target genes involved in different cellular processes.

(19)

19

Fig. 2.3. MiRNA biogenesis [50]. 2.2.2. MiRNA–mRNA interactions

MiRNAs function by targeting complementary sequences in mRNA transcripts, usually in the 3′ untranslated region (3′ UTR), and prevent protein synthesis by inhibiting translation or inducing target-gene degradation. The seed sequence of miRNA (the first 2–8 nucleotides starting at the 5′- end and counting toward the 3′-end) has to be perfectly complementary to target mRNA sequence in order to form Watson-Crick pairs with the cognate target-gene (Fig. 2.4). A Watson-Crick match between a miRNA and mRNA nucleotide occurs when adenosine (A) pairs with uracil (U) and guanine (G) pairs with cytosine (C). G:U wobble in the seed match refers to the allowance of a G pairing with a U instead of a C. Furthermore, mRNA can be repressed by more than one miRNA species at the same time. The level of repression achieved is dependent on both the amount of mRNA and the amount of available miRNA complexes [51].

(20)

20

Fig. 2.4. Schematic miRNA:mRNA interaction. MiRNA position number is

shown in blue. The seed sequence refers to nucleotides in miRNA position number 2–8. Flank refers to the mRNA sequence on either side of the region

corresponding to the miRNA seed sequence. Watson-Crick matches in the seed sequence are shown in red, and an example of G-U wobble in the seed

sequence is shown in green [47].

Discovering miRNA expression patterns that are unique to a particular cancer and their target-genes is an important step in identifying biomarker signatures that will be effective in cancer detection [52].

2.2.3. Single nucleotide polymorphisms influence miRNA function

Since the discovery of miRNAs, a great number of studies had character-rized genetic polymorphisms in miRNA genes and genes, encoding proteins important in miRNA processing, that affect miRNA regulation [53], affect the miRNA–mRNA interaction and influence the development of diseases [9], especially cancer. However, sequencing has shown that single nucleotide polymorphisms (SNPs) in miRNA coding genes and specifically in miRNA seed regions are rare [57, 58].

SNPs in miRNA genes are thought to affect miRNA processing and function in three ways [53, 59]: (i) through the transcription of the primary transcript (SNPs in DROSHA, DGCR8 and etc.). For example, Dicer was found to be reduced in GC tissues in both mRNA and protein levels and down-regulation of Dicer was correlated with tumor differentiation and lymph node invasion in GC tissues, which suggested an essential role of Dicer in cancer invasion [57]. (ii) through pri-miRNA and pre-miRNA processing (SNPs in pri-, pre-miRNA). SNP in hsa-miR-27a gene (rs11671784) was shown to be associated with GC [9] and SNP in pri-miR-365b (rs121224) was shown to be associated with intestinal-type GC [58]. SNPs can occur in the pri-miRNA and pre-miRNA strands and are likely to affect miRNA processing and subsequent mature miRNA levels. Such SNPs can lead to either an increase or decrease in processing [50]. (iii) through effects on miRNA–mRNA interactions (SNPs in

miRNA seed and in mRNA regulatory regions) [8, 53] (Fig. 2.5). SNPs in

(21)

21

modulate the miRNA–mRNA interaction and create or destroy miRNA binding sites [19–21, 53].Target site polymorphisms still remain poorly explored in different cancers including GC. The importance of miRNA related SNPs in gene regulation and the mechanism by which these SNPs can induce alteration in molecular pathways is largely unknown.

Fig. 2.5. Example of SNPs in miRNA seed and mRNA regulatory regions. 2.2.4. Role of miRNAs in atrophic gastritis

Gastric carcinogenesis is a multifactorial H. pylori-triggered dynamic process that goes through a cascade of preneoplastic conditions [59]. Important studies have shown that miRNAs are already deregulated in early stages of gastric carcinogenesis, including H. pylori induced gastritis and premalignant stages of gastric atrophy and intestinal metaplasia. Study by Petrocca et al. showed that chronic inflammation of the gastric mucosa was associated with alteration of seven miRNAs, specifically hsa-miR-155 [60].

(22)

22

In the recent study by Link et al. the expression of hsa-miR-155, as well as hsa-miR-223 was shown to be increased in chronic non-atrophic gastritis (CNAG) and atrophic gastritis (AG) compared with normal H. pylori-nega-tive gastric mucosa [6]. Matsushima et al. profiled tissue samples from

H. pylori positive and negative subjects and also described an upregulation of

hsa-miR-223, as well as down-regulation of other 30 miRNAs (such as has-let-7, hsa-miR-101, hsa-miR-103, hsa-miR-204, hsa-miR-223, hsa-miR-377, etc.). Moreover, eight of the miRNAs (miR-223, miR-200a, 31, has-let-7e, 141, 203, 204 and hsa-miR-455) enabled discrimination of H. pylori status with acceptable accuracy [61]. These miRNAs were shown to be related with cancer: Kang W et al. confir-med oncogenic role of hsa-miR-223 by targeting STMN1 gene in GC [62], whereas Li Bo-sheng et al., validated hsa-miR-223 expression in plasma of GC patients [63]. MiRNAs previously shown to be associated with GC, were also shown to be deregulated in precancerous conditions, e.g. hsa-miR-200a was shown to be upregulated in the GC tissue [64]. Other study summarized that hsa-miR-200 family, consisting of 5 members (hsa-miR-200a, -200b, -200c, -141, -429), was associated with cancer initiation and metastasis [65]. This confirms that a number of miRNAs, which are associated with GC are abnormal in precancerous stages as well, so those miRNAs could be used for early GC diagnostic.

2.2.5. Role of miRNAs in gastric cancer

An increasing number of studies have identified miRNAs as potential biomarkers for GC diagnosis, prognosis and therapeutic targets [7, 69, 70]. Volinia et al. have provided one of the first miRNA expression profiles in GC tissue showing a specific deregulation pattern [14]. In tumors up-regulated miRNAs inhibit tumor suppressors, leading to cell proliferation, invasion and reduced apoptosis. These miRNAs are named oncoMiRs. In contrast, down-regulated miRNAs target oncogenes and facilitate the activity of their target oncogenes. Many studies have reported significant deregulation of miRNAs belonging to 17, 19, 21, 223, miR-135 and other families. Rewiev by Wadhwa [68] indicated the members of hsa-miR-17-92 cluster (hsa-miR-17, hsa-miR-18a, hsa-miR-19a, hsa-miR-20a, hsa-miR-19b-1 and hsa-miR-92) to play an important role in the oncogenesis of GC by inhibiting tumor suppressor genes. Hsa-miR-17 may function as oncogene and promote cancer development by negatively regulating tumor suppressor genes or genes that control cell proliferation [68]. Hsa-miR-19 consists of three subclasses in humans: miR-19a, miR-19b1 and hsa-miR-19b2. hsa-miR-19 has been shown to downregulate phosphatase and

(23)

23

tensin homolog (PTEN) and effectively increase activity of the cellular survival-promoting signal pathway PI3K-Akt [69]. Another oncomiR - hsa-miR-21 has been found to be deregulated in a wide variety of human cancers, including breast, colorectal, lung, pancreas, skin, liver, gastric, cervical, thy-roid, various lymphatic and hematopoietic cancers. Interestingly, high levels of hsa-miR-21 may not only characterize cancer cells but also represent a common feature of pathological cell growth or cell stress [70]. Hsa-miR-155 is a commonly up-regulated oncomiR in human cancers and is suggested as a novel non-invasive biomarker for human cancer detection [71]. Among reported studies in GC tissues, hsa-miR-21, hsa-miR-25, hsa-miR-92, hsa-miR-223 were the most consistently up-regulated miRNAs, while miR-375 and hsa-miR-148 were the most consistently down-regulated [15–17].

Interestingly, some of miRNAs have opposite deregulation directions in the presence of GC, as separate profiling studies show significant inconsistency among deregulated miRNAs [72]. Hsa-miR-9 was found to be up-regulated in two studies [16, 76], while two other papers showed signifycant down- regulation of this miRNA in GC tissues [63, 77]. These discrepancies regarding opposite direction of deregulation are most likely linked to differrences in anatomical location of the GC, histological subtype, disease stage, profiling platforms, statistical analysis and many other potential confounding factors. Besides, the majority of currently published studies on miRNA profile in GC tissues cover subjects from Asian countries; meanwhile, data on European GC patients is still scarce [76].

Based on the literature survey one could point out a huge number of miRNAs or miRNA families to be associated with GC. However, the acquired results in GC miRNA research are conflicting, therefore further studies are needed in order to validate them.

2.2.6. Circulating miRNAs

MiRNA presence in blood was discovered in 2008. They can be detected in plasma and blood cells (platelets, erythrocytes and nucleated blood cells) [55, 79, 80]. Their occurrence in plasma or serum maybe the result of cell death inside a tissue, or secretion by certain cells in response to a stimulus [81, 82]. However, the origin of these miRNAs is not completely determined. In addition, they can also be secreted in a free form by some other mechanisms [82, 79]. For example, food-derived plant miRNAs may pass through the gastrointestinal (GI) tract, enter into the plasma and serum and interact with mammalian endogenous RNAs and regulate their expression [83, 84]. Furthermore, miRNAs are stable and resistant to pH variations, long-term storage at room temperature, and repeated freeze–thaw cycles both in serum

(24)

24

and plasma samples [55, 85, 81]. It is assumed that the stability of circulating miRNAs and their high resistance to degradation by blood nucleases is due to miRNA complexes with various proteins, such as Ago2, lipoproteins, exosomes and other microparticles in blood (Fig. 2.6).

Since the discovery of circulating miRNAs in body fluids, an increasing number of studies have focused on their potential as non-invasive biomarkers for many diseases, particularly for cancers [55, 86]. Review by Schwarzen-bach et al. summarizes clinical relevance of circulating miRNAs and provides lists of cell-free miRNAs that can be used as prognostic and predictive biomarkers in cancer and discusses their utility in personalized medicine. For example, hsa-miR-20a, hsa-miR-24, hsa-miR-25, hsa-miR-145, hsa-miR-152, hsa-miR-199-5p, hsa-miR-221, hsa-miR-222, hsa-miR-223, hsa-miR-320 circulating in serum or plasma may be used as non-invasive biomarkers for the detection of lung cancer, hsa-miR-200b for breast cancer,hsa-miR-375, hsa-miR-141, hsa-miR-378*, hsa-miR-409-3p for prostate cancer and so on [85].

Fig. 2.6. Putative sources of circulating miRNAs.

(25)

25

2.2.7. Circulating miRNAs in gastric cancer

Studies indicate that there are less prognostic miRNA biomarkers in blood compared to tissue of GC patients [86]. Two possible reasons have been proposed: first, there are more studies investigating miRNAs in gastric tissues; second, miRNAs are selectively released into blood. Furthermore, studies indicate that circulating miRNAs are not GC phenotype specific, i.e. miRNAs in blood significantly correlate with no more than two clinicopatho-logical features, whereas miRNAs in tissues are associated with more features. For example, Liu et at al. identified five miRNAs (hsa-miR-1, hsa-miR-20a, hsa-miR-27a, hsa-miR-34, hsa-miR-423-5p) as biomarkers for GC detection and the expression level of these miRNAs correlated with tumor stage [87]. However, miRNAs in blood and tissues of GC patients are associated with survival, i.e. most of the miRNAs with high expression level were detected in the poor survival groups of GC patients. For example, study by Zhu et al. identified and validated five differentially expressed miRNAs (hsa-miR-16, hsa-miR-25, hsa-miR-92a, hsa-miR-451 and hsa-miR-486-5p) in plasma of patients diagnosed with gastric non-cardia adenocarcinoma (GNCA). More importantly, these five miRNAs showed high diagnostic performance in the group of patients with early-stage of GNCA [88].

(26)

26

3. MATERIALS AND METHODS

3.1. Ethics statement

The study was approved by the Kaunas Regional Biomedical Research Ethics Committee (Protocol No BE-2-10). Written informed consent was obtained from all study participants.

For Study I, patients and controls were recruited during the years 2005– 2013 at three gastroenterology centers in Germany (Department of Gastroen-terology, Hepatology and Infectious Diseases, Otto-von-Guericke University, Magdeburg), Lithuania (Department of Gastroenterology, Lithuanian Univer-sity of Health Sciences, Kaunas) and Latvia (Riga East UniverUniver-sity Hospital and Digestive Diseases Centre GASTRO, Riga). Controls were patients from the out-patient departments, who were referred for upper endoscopy because of dyspeptic symptoms and had no history of previous malignancy. GC patients had histological verification of gastric adenocarcinoma and were recruited from out-patient and stationary departments.

For study II patients and controls were included during the years 2007– 2015 at Department of Gastroenterology and Department of Surgery, Lithua-nian University of Health Sciences (Kaunas, Lithuania). Control group consis-ted of patients who had no history of previous malignancy. All patients had histological verification of diagnosis. All patients signed an informed consent form to participate in the study.

3.2. Major reagents

TaqMan MiRNA Reverse Transcription Kit and miRNA-specific RT stem-loop primers, High Capacity Reverse Transcription Kit, TaqMan MiRNAs Assay, TaqMan® Low Density Array Card, TaqMan Universal PCR Master Mix, inhibitor miR-20b-5p and mimics miR-451a-5p and hsa-miR-1468-5p and nonspecific control (NC) were purchased from Thermo Fisher Scientific (former Applied Biosystems).

The miRNeasy Micro Kit, RNease Mini Kit, RNease Micro Kit, miRNeasy Serum/ Plasma Kit were purchased from Qiagen (USA).

ELISA kit to detect the serum-specific IgG antigen was purchased from Virion/Serion GmbH (Germany).

RNAlater, mirVana miRNA Isolation Kit was purchased from Thermo Fisher Scientific (former Ambion, Austin (USA)).

Human gastric carcinoma cell line AGS (CRL-1739™) was obtained from American Cell Bank (ATCCTM), GC cell line MKN-28 was provided by

(27)

27

Alexander Link (from Department of Gastroenterology, Hepatology and Infectious Diseases, Otto-von-Guericke University (Magdeburg, Germany).

Fetal bovine serum, Hams’s F-12K (Kaighn’s) medium, penicillin and streptomycin (5,000 U/ml) were purchased from Gibco™ by Life Techno-logies (USA). Lipofectamine 3000 were purchased from Thermo Fisher Scientific (former Invitrogen (USA)). Trypsin and phosphate-buffered saline (PBS) were purchased from Sigma-Aldrich (USA).

3.3. Study I

3.3.1. Study design and patients samples

The aim of Study I was to evaluate associations between miRNA target-genes IL12B, INSR, CCND1 and IL10 polymorphisms and GC in European population. In total, 982 individuals were included in this study (508 controls and 474 GC). There were 206 subjects from Germany (104 controls and 102 GC), 285 subjects from Latvia (146 controls and 139 GC) and 491 subjects from Lithuania (258 controls and 233 GC). All patients were of European descent. Clinical characteristic and detailed description of patients is presented in the article “Polymorphisms of microRNA target-genes IL12B, INSR, CCND1

and IL10 in gastric cancer”.

3.3.2. DNA extraction, selection of putative miRNA target-gene single nucleotide polymorphisms and genotyping

Genomic DNA was extracted using salting out method from peripheral blood mononuclear cells [89]. In order to select the candidate SNPs falling within 3’-UTR of genes, which are putative targets of frequently deregulated miRNAs in GC, the mirsnpscore database was used (http://www.bigr.medisin. ntnu.no/mirsnpscore). Using this bioinformatic approach we selected four genes: IL12B (rs1368439), INSR (rs1051690), CCND1 (rs7177) and IL10 (rs3024498) that have potential target sites of miR-27, miR-146a, hsa-miR-223 and hsa-miR-107. IL12B T>G (rs1368439), INSR T>C (rs1051690),

CCND1 A>C (rs7177) and IL10 T>C (rs3024498) SNPs were genotyped using

TaqMan® genotyping method. A detailed description is presented in the publication “Polymorphisms of microRNA target-genes IL12B, INSR, CCND1

(28)

28 3.4. Study II

3.4.1. Study design

First (Part I), miRNA profile was determined in GC tissues compared to the normal healthy gastric mucosa. Later, expression level of selected miRNAs was validated in GC tissue and plasma samples of GC patients and healthy controls.

Second (Part II), miRNA profile was determined in AG and GC tissue compared to the normal healthy gastric mucosa. In replication analysis we validated profiling results, see Table 3.1.

Table 3.1. Design of study II

Design of Study II Type of tissue

Number of samples

Number of

miRNA review Used method

Part I Profiling cohort Mucosa of stomach GC, n = 13

HC, n = 12 377 qRT-PCR with TLDA card A Validation cohort Mucosa of stomach GC, n = 38 HC, n = 39 6 qRT-PCR with TaqMan assay Plasma 4 Part II Profiling cohort Mucosa of stomach GC, n = 20 AG, n = 18 HC, n = 26 2,180 Small RNA-seq Validation cohort Mucosa of stomach GC, n = 39 AG, n = 40 HC, n = 40 21 qRT-PCR with custom TLDA

GC – gastric cancer, AG – atrophic gastritis HC – healthy control, TLDA – TaqMan low density array.

(29)

29

3.4.2. Study population of part I of study II

The study included a total of 51 control subjects and 51 GC patients, who were divided into the profiling (GC, n = 13; controls, n = 12) and validation cohorts (GC, n = 38; controls, n = 39). Gastric tissues (obtained during endoscopic or surgical procedures) and plasma samples were collected from healthy patients and patients with GC. Detailed description of study population used for this part of the study is presented in the article “Analysis of

Deregu-lated microRNAs and Their Target-genes in Gastric Cancer”.

3.4.3. Study population of part II of study II

The study included a total of 59 control subjects, 58 AG patients and 66 GC patients, which were divided into the profiling (GC, n = 20; AG, n = 18 and controls, n = 26) and validation (GC, n = 39; AG, n = 40 and controls, n = 40) cohorts. Clinical and pathological characteristics of the patient cohorts including age, gender and disease stage are summarized in Table 3.2. Profiling analysis was performed using next-generation sequencing and validation analysis was carried out using custom RT-qPCR with TLDA custom card of selected miRNAs in the larger cohort.

(30)

30

Table 3.2. Clinical characteristics of the gastric cancer and atrophic gastritis patients and healthy controls.

Profiling cobort (n = 64) Validation cohort (n = 119)

p-value Gastric cancer (n = 20) Atrophic gastritis (n = 18) Healthy control (n = 26) Gastric cancer (n = 39) Atrophic gastritis (n=40) Healthy control (n=40) Age Mean± 65 ± 10.9 68.9 ± 8.9 60.9 ± 15 70.5 ± 10.8 63.3 ± 11.7 57 ± 16.5 Gender Male, n (%) 15 (75) 3 (16.7) 7 (26.9) 24 (61.5) 16 (40) 10 (25) > 0.05* Female, n (%) 5 (25) 15 (83.3) 19(73.1) 15 (38.5) 24 (60) 30 (75) Lauren classification Diffuse, n (%) 10 (50) 15 (38.5) > 0.05* Intestinal, n (%) 10 (50) 24 (61.5) Mixed, n (%) 0 0 Unknown, n (%) 0 0 H. pylori infection Positive, n (%) 9 (45) 7 (38.9) 11 (42.3) 24 (61.5) 29 (72.5) 20 (50) _{>0.05 [GC]} [Con] 0.02 [AG]* Negative, n (%) 6 (30) 11 (61.1) 15 (57.7) 9 (23) 11 (27.5) 18 (45) Unknown, n (%) 5 (25) 0 0 6 (15.4) 0 2 (5) Tumor localization Cardia, n (%) 2 (10) 6 (15.4) >0.05* Corpus, n (%) 9 (45) 22 (56.4) Antrum, n (%) 9 (45) 8 (20.5) Linitis plastica, n (%) 0 3 (7.7) TNM staging I, n (%) 6 (30) 7 (18.0) 0.02* II, n (%) 2 (10) 11 (28.2) III, n (%) 10 (50) 7 (17.9) IV, n (%) 2 (10) 12 (30.8) Unknown, n (%) 2 (5.1)

(31)

31 Table 3.2. Continued

Profiling cobort (n = 64) Validation cohort (n = 119)

p-value Gastric cancer (n = 20) Atrophic gastritis (n = 18) Healthy control (n = 26) Gastric cancer (n = 39) Atrophic gastritis (n=40) Healthy control (n=40) T 1/2 8 (40) 12 (30.8) > 0.05* 3 7 (35) 12 (30.8) 4 5 (25) 13 (33.3) Unknown, n (%) 0 2 (5.1) N 0 9 (945) 14 (35.9) > 0.05* 1 3 (15) 12 (30.8) 2 3 (15) 6 (15.4) 3 5 (25) 5 (12.8) Unknown, n (%) 0 2 (5.1) M 0 14 (70) 14 (35.9) 0.05 * 1 2 (10) 12 (30.8) Unknown, n (%) 4 (20) 13 (33.3) Differentiation grade 1/2 10 (50) 12 (30.8) > 0.05* 3 10 (50) 21 (53.8) Unknown, n (%) 0 6 (15.4)

* – p value meaning is calculated compared profiling and validation cohort in each group (GC to GC; AG to AG and HC to HC groups). p > 0.05 means that groups are not statistically different, (Mann-Whitney test).

(32)

32

3.4.4. Tissue sample preparation and RNA extraction

Gastric biopsy samples were obtained from antral part of the stomach from control subjects who were referred for upper GI endoscopy due to dyspeptic symptoms and had no previous history of malignancy. GC tissue specimens were obtained from surgical specimens immediately after resection from patients undergoing primary surgery for GC with no preoperative irradiation and chemotherapy. Gastric adenocarcinoma in GC patients was verified by histology and classified according to Lauren into diffuse and intestinal types [25]. H. pylori status was assessed in GC, AG and control subjects using indirect ELISA to detect the serum-specific IgG antigen. Gastric tissue samples were stored in RNAlater for +4°C and 24 hours and later stored at 80°C. 30 mg of tissue was homogenized in sterile conditions before total RNA isolation with mirVana miRNA Isolation Kit for profiling study (part I of study II) and miRNeasy Mini Kit for profiling study (part II of study II) and validation studies (part I of study II and part II of study II), according to the manufacturers’ instruction. The quality and quantity of RNA was assessed using the Nanodrop 2000 spectrophotometer (Thermo Scientific, USA).

3.4.5. Plasma sample preparation and RNA extraction

Plasma samples were collected from healthy patients and patients with GC. MiRNA profiling (part I of study II) was carried out using plasma samples from 38 GC patients and 39 healthy volunteers. Venous blood was collected in EDTA anticoagulation vacuum tubes and was centrifuged at 3500 rpm for 15 min at room temperature; separated plasma was transferred into 1.5 ml tubes and placed in a −80°C freezer for short-term storage. Small RNAs were extracted from 200 μl of plasma using the miRNeasy Serum/ Plasma Kit according to the manufacturers’ instruction.

3.4.6. MicroRNA profiling using the TaqMan Low Density Array card A

MiRNA profiling (part I of study II) in GC tissue compared to normal gastric tissue was carried out using TLDA card A which enabled to quantify 374 miRNAs. Complementary DNA (cDNA) was reverse transcribed from total RNA samples using MegaplexTM pool. More details are presented in publication “Analysis of Deregulated microRNAs and Their Target Genes in

(33)

33

3.4.7. MicroRNA validation using the custom TaqMan Low Density Array

Validation of miRNA expression profiles (part II of study II) was perfor-med in GC, AG and normal tissue using TLDA custom card which enabled to quantify 24 selected miRNAs. In short, 600 ng of total RNA was initially reverse transcribed using Custom RT pool (G1406). 0.88 μl of cDNA, 49.9 μl of TaqMan® Universal Master Mix II, No AmpErase UNG (2X) and 49.1 μl nuclease free water was loaded per port on 8-port TLDA card and run on ViiA 7 Real-time PCR System (Applied Biosystems) following the standard pro-tocol. All GC and AG tissue samples were randomized and placed with healthy tissue controls on the same TLDA card. The expression level of miRNAs in tissue was normalized to the mean expression values of hsa-miR-16-5p.

3.4.8. Quantitative real-time polymerase chain reaction with TaqMan assay

TaqMan® miRNA assays were used to quantify tissue-validated miRNAs in tissue and plasma samples (part I of study II). Reverse transcription was performed using TaqMan® MiRNA Reverse Transcription Kit and miRNA-specific RT stem-loop primers. Singleplex qPCR reactions were scaled down to 7.5 μl and performed according to the manufacturer's instructions. Each sample was run in duplicate. The expression level of miRNAs in tissue and plasma was normalized to the mean expression values of miR-127 and hsa-miR-16-5p, respectively [90,91].

3.4.9. Preparation of small RNA libraries and next generation sequencing

Small RNA libraries were prepared using TruSeq Small RNA Sample Preparation Kit (Illumina) according to the manufacturer's protocol with 1μg RNA input per sample followed by RNA 3' adapter ligation, RNA 5' adapter ligation, cDNA synthesis, PCR amplification using unique barcode sequences for each sample and gel size-selection of small RNA library. The yield of sequencing libraries was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies). Small RNA libraries were pooled with 6 samples per lane for plasma and 24 samples per lane for tissue samples and sequenced on HiSeq2500 (Illumina) next-generation sequencing platform.

(34)

34

3.4.10. Small RNA-seq data and statistical analysis

Statistical analysis of study II part I is presented in publication “Analysis of Deregulated microRNAs and Their Target Genes in Gastric Cancer”. In study II part II in profiling cohort several filtering steps were performed after obtaining and multiplexing the raw reads. At first, cutadapt was used to trim low-quality ends of the reads with Phred quality score value (Q) > 30 and to remove 3' adapter (5’-TGGAATTCTCGGGTGCCAAGG-3’) sequences from the reads. The trimmed sequences shorter than 17 bp were discarded. Second, in order to reduce computational time the reads with exact sequences were collapsed while saving the information about the read counts. Third, the collapsed reads which mapped to viral genome (RefSeq) [92], human tRNAs, rRNAs, snRNAs, sRNAs (Rfam) [93] and non-human miRNA precursors (miRBase v20) [94] were filtered using BLASTN v2.2.30 [95]. The mature miRNA quantification in filtered reads were performed using the quantifier.pl module from miRDeep2 v2.0.0.7 software [96], with default parameters using mature miRNA sequences as reference (miRBase v20). Differential expression analyses of the size factor normalized counts of mature miRNAs between samples were performed by use of negative binomial generalized linear models implemented DESeq2 package [97]. The p-values resulting from Wald tests [98] were corrected for multiple testing using the false discovery rate (FDR) proposed by Benjamini and Hochberg [99]. The mature miRNAs with a corrected p < 0.01, and fold change > 2 were considered to be significantly differentially expressed. Power of the miRNA-seq study was 80% in GC group and 99% in AG group (Fig. 3.1).

(35)

35

Fig. 3.1. Power of miRNA-seq study in gastric cancer (A) and in

atrophic gastritis (B). Effect size, when: (i) fold change = 2 (p < 0.01) showed in red, (ii) fold change = 3 (p < 0.01) showed in blue,

(iii) fold change = 4 (p < 0.01) showed in green.

The differentially expressed miRNAs were selected for validation study in the following steps: (i) differentially expressed miRNAs in tissue were filtered based on their expression level in plasma (read count sum of miRNAs present in plasma samples > 5); (ii) in order to show similarities in miRNA expression between GC and AG conditions, overlapping differentially expressed miRNAs in both conditions were selected when compared to HC; (iii) top differentially expressed miRNAs (lowest FDR adjusted p-value and highest fold change value) were chosen for both GC and AG conditions in order to select the most informative and condition-specific miRNAs.

In order to evaluate the prediction performance of GC and AG miRNA expression profiles in plasma, radial support vector machine (SVM) and logistic regression models were built. The samples containing normalized miRNA expression data were used to construct the model. The tuning parame-ters thereby were trained using 10-fold cross-validation. For model testing the samples were randomly split into two sets at 5:3 ratio and the smaller partition was used to evaluate models prediction performances. The classifiers’ perfor-mances were measured by the area under the receiver operating characteristic (ROC) curve, sensitivity (SN) and specificity (SP). The statistical prediction analysis was performed using caret [100] and ROCR [101] packages.

(36)

36 3.5. Study III

3.5.1. Study design

First, we performed 20b-5p, 451a-5p and hsa-miR-1468-5p expression analysis in GC cells (AGS and MKN-28) compared to normal tissue. Second, putative target-genes of these miRNAs were selected using mathematics algorisms in “R” package. Third, GC cells transfection with inhibitor of hsa-miR-20b-5p and mimics hsa-miR-451a-5p and hsa-miR-1468-5p was performed using Lipofectamine 3000. Finally, putative target-gene expression was performed 24 h and 48 h after transfection using qRT-PCR.

3.5.2. Evaluation of miRNAs target-genes

MiRNAs target-genes were evaluated using mathematical algorithms in “R” package (RStudio Desktop v. 1.0.143) software. First, DisGenet data base was used to determine GC associated genes. Second, DIANA-microT-CDS, ElMMo, MicroCosm, miRanda, miRDB, PicTar, PITA and TargetScan mathematical prediction tools were used for hsa-miR-20b-5p, hsa-miR-451a-5p, hsa-miR-1468-5p potential target-genes analysis. Next, depending on information in miRecords, miRTarBase and TarBase target-genes, which were already validated in other studies, were rejected. Finally, target-genes were filtrated according their function in oncogenesis (oncosupressor or protoon-cogene) and literature analysis was performed.

3.5.3. Cell culture conditions

The AGS and MKN-28 cells were cultured in Ham’s F-12K medium containing 10% fetal bovine serum, 1% penicillin-streptomycin solution and incubated at 37 °C, 5% CO2 and saturated humidity. Cell growth was observed under an inverted microscope. When cell growth reached 70% to 80% con-fluence, the cells were digested with 0.25% trypsin and passaged. The culture medium was changed every other day, and the cells were passaged every 3 to 4 days. Cells in the logarithmic growth phase were collected for experiments.

3.5.4. MiRNA genes expression analysis in gastric cancer cells

Total RNA from gastric cells AGS (n = 7) and MKN-28 (n = 7) and normal gastric tissue (n = 11) was extracted using miRNeasy Micro Kit in accordance with manufacturer’s recommendations. Complementary DNA was reverse transcribed from total RNA samples using TaqMan miRNA assays miR-20b (001014), miR-451a (001141), miR-1468 (121107_mat) and hsa-miR-16 (000391) as endogenous control. The resulting cDNA was subsquently

(37)

37

amplified by PCR using TaqMan assay and Universal Master Mix according to the manufacturer's recommendations using a qRT-PCR System ABI 7500 Fast (Applied biosystem, USA).

3.5.5. Gastric cancer cells transfection

GC cells (AGS and MKN-28) were seeded into twenty-four-well culture plates at a concentration of 40,000 cells/well. The volume in each well was 500 μl. Once the cells adhered, they were transfected with the inhibitor of hsa-miR-20b-5p, mimics of hsa-miR-451a-5p and hsa-miR-1468-5p, and nonspecific control (NC) according to the Lipofectamine 3000 transfection manual. Mimic positive control (PC) (hsa-miR-1) and inhibitor PC (hsa-anti-let-7c) were used to optimize transfection experiments. Transfection procedure was performed three times with different passages of GC cells.

3.5.6. MiRNA putative gene expression analysis in gastric cancer cells

Total RNA was extracted from transfected cells (AGS and MKN-28) using the RNease Micro Kit and reverse transcription was performed from 1 µg of total RNA in 20 µl reaction using High-Capacity Reverse Transcription Kit in accordance to manufacturer’s recommendations. The mRNA expression levels of the putative target-genes EREG, FAT4, IRF1, TXNIP, PTEN for hsa-miR-20b; CAV1 and ADAM28 for hsa-miR-451a; TNFα, DNMT1 and CITED2 for hsa-miR-1468 were assessed by quantitative RT-PCR 24 h and 48 h after transfection, using TaqMan Gene Expression assays. Two technical replicates were made for each gene assay and reaction was performed on ABI 7500 Fast Real-Time PCR system (Applied Biosystems, USA). Relative fold mRNA levels were determined using the 2−ΔΔCt method, with ACTB as an internal control.

3.5.7. Statistics

The data are expressed as the mean dCt of three independent experiments and processed by GraphPad Prism software (GraphPad Software Inc., San Diego, USA). The Mann Whitney U-test were used to distinguish differences between groups. A P value of < 0.05 was considered statistically significant.

(38)

38

4. RESULTS

4.1. Study I

Similar distribution of genotype and allele frequencies was observed between GC patients and controls for all polymorphisms under study (IL12B (rs1368439), CCND1 (rs7177) and IL10 (rs3024498)) except for INSR SNP (rs1051690). The frequency of the rare allele T of INSR gene was significantly higher in GC patients than in controls (23.26% and 19.19% respectively, P=0.028). CT genotype was also more prevalent in patients compared to cont-rol group (38.48% and 30.12% respectively, p < 0.021). Logistic regression analysis revealed that only one polymorphism (rs1051690 in INSR gene) was associated with increased risk of GC. Carriers of CT genotype had higher odds of GC when compared to CC genotype (OR – 1.45, 95% PI 1.08 – 1.95, P = 0.01). Similar association was observed in a dominant model for INSR gene, where combination of TT+CT genotypes showed an increased risk for GC (OR – 1.44, 95% PI 1.08 – 1.90, P = 0.01). Other analyzed SNPs were not associated with the presence of GC. A detailed description of the study results

is presented in the article “Polymorphisms of microRNA target genes IL12B, INSR, CCND1 and IL10 in gastric cancer”

4.2. Study II, Part I

4.2.1. MiRNA expression profiling of GC and healthy gastric mucosa tissue by Taq-Man low-density array

Tissue miRNA profiles from GC patients were compared with profiles of gastric mucosa from healthy individuals and 15 differentially expressed miRNAs were identified, 7 of which were up-regulated and 8 were down-regu-lated with FDR adjusted p-value < 0.01 and fold change > 2. Among these, 6 miRNAs (135a-5p, 148a-3p, 204-5p, hsa-miR-375, hsa-miR-223-3p and hsa-miR-155-5p), showing the highest significance and fold change values, were selected for the validation study. A detailed

description of the study results is presented in the article “Analysis of Deregulated microRNAs and Their Target Genes in Gastric Cancer”.

(39)

39

4.2.2. MiRNA validation in gastric cancer and healthy gastric mucosa tissue with quantitative real-time polymerase chain reaction

The six candidate miRNAs selected for verification were quantified using qRT-PCR. The expression levels of hsa-miR-148a-3p, hsa-miR-204-5p, hsa-miR-223-3p and hsa-miR-375 in GC tissue showed significant differ-rential expression in the same direction as in the discovery study (FDR adjusted p < 0.05). A detailed description is presented in the article “Analysis

of Deregulated microRNAs and Their Target Genes in Gastric Cancer”.

4.2.3. Plasma miRNAs as potential biomarkers for gastric cancer

In GC tissue validated hsa-miR-148a-3p, hsa-miR-204-5p, hsa-miR-223-3p and hsa-miR-375 were further analyzed in plasma samples. The relative levels of the candidate miRNAs in individual samples were determined using qRT-PCR. The hsa-miR-16-5p was used as the endogenous control to normalize the expression levels of candidate miRNAs. The expression levels of hsa-miR-148a-3p and hsa-miR-375 showed significant down-regulation in GC plasma; while hsa-miR-223-3p was significantly up-regulated (FDR adjusted p < 0.05). No significant differences were detected in the expression levels of hsa-miR-204-5p. A detailed description is presented in the article

“Analysis of Deregulated microRNAs and Their Target Genes in Gastric Cancer”.

4.3. Study II, part II

4.3.1. Global overview of miRNA transcriptome

Small RNA-seq of normal stomach tissue, AG and GC tissue samples yielded ~ 271M raw sequencing reads (from ~417K to ~29.5M reads/sample). Pre-filtering and filtering steps retained 50.4% (~137M) of initial raw reads. The majority of filtered reads were of 20–23 nt length which corresponds to the range of mature miRNA sequences. Quantification of filtered reads and identification of known miRNAs have yielded ~56.5M sequences to be mapped to 1,661 known miRNAs from miRBase v20.

(40)

40

4.3.2. MiRNA expression profiling of gastric cancer,

atrophic gastritis and healthy gastric mucosa by small RNA-seq

The similarity structure of the tissue miRNA transcriptomes among GC, GCadj, AG and HC was identified by a multidimensional scaling (MDS) analysis using Spearman's correlation distance (1-correlation coefficient). In the picture we can see four different groups, and it is interesting that GCadj and HC show miRNA expression profile similarity (Fig. 4.1).

Fig. 4.1. Similarity structure of miRNA transcriptome generated by next

generation sequencing. Multidimensional scaling plot showing four clusters corresponding to tissue of gastric cancer (GC), gastric cancer adjacent (GC

adj), atrophic gastritis (AG) and normal mucosa (healthy control (HC)). The analysis was performed on normalized miRNA count data using

Spearman's correlation distance (1-correlation coefficient).

For identification of differentially expressed miRNAs among cancerous, precancerous and healthy gastric mucosa tissue specimens, the comparative analysis of normalized small RNA-seq expression data was performed.

ROLE OF MICRORNA AND THEIR TARGET GENES IN PATHOGENESIS OF GASTRIC CANCER

Violeta Šaltenienė

ROLE OF MICRORNA AND THEIR

TARGET GENES IN PATHOGENESIS

OF GASTRIC CANCER

Violeta Šaltenienė

MIKRO RNR IR JŲ REGULIUOJAMŲ

GENŲ VAIDMUO SKRANDŽIO

VĖŽIO PATOGENEZĖJE

CONTENTS

ABBREVIATIONS

INTRODUCTION

1. AIM AND OBJECTIVES

2. REVIEW OF LITERATURE

3. MATERIALS AND METHODS

4. RESULTS